ANALYSIS
The LCP_Empire markdown script shows how to download SRTM raster data covering the extent of the Roman Empire, mosaic the tiles together and process the elevations to generate a conductance matrix with gdistance package functions. This matrix is a useful input for calculations of travel time between point coordinates in the raster, mimicking discrete source-destination travel by foot. The walking function can be altered to other travelling modes, howver the main intent is to model for terrestial travel between settled areas in the Roman Empire. In the exemplar, Hanson's list of cities is filtered for sites with estimated population over 1000 to generate time travel rasters, which can produce least cost paths accounting for terrain friction.
- Adela Sobotkova
, SDAM project, admin@ancientsocialcomplexity.org
- [Name], [ORCID], [Institution], [email]
- [Name], [ORCID], [Institution], [email]
- [Name], [ORCID], [Institution], [email]
CC-BY-SA 4.0, see attached License.md
[Here will be DOI or some other identifier once we have it]
[Here will go related articles or other sources we will publish/create]
The first example uses only a single SRTM tile and cities in Italy for calculations. A loop is provided to repeat analysis at this scale to deliberate part of the region. The full example downloads some 50 SRTM tiles, and takes a couple hours to run. The final distance calculations run for empire-wide urban footwalks takes days.
The LCP_Empire.Rmd was created and run on Ucloud, using a 192Gb RAM and unlimited storage. The final mosaiced raster has 158.4Mb and the conductance transition layer takes up 2.1 Gb of memory in R (928.9 Mb file storage). The conductance matrix is stored on sciencedata.dk > SDAM_data>landscape folder. Choose an adequately resourced machine or descale or alter the computational tasks. One can reduce the spatial extent - currently the southern boundary is deep in Sahara, which is counterproductive. One can also further aggregate the mosaiced raster by a factor of 10.
The scripts reply on R libraries of tidyverse, raster, sf, and gdistance. Depending on where the point data comes from, you may also need jsonlite and getPass or sdam library.
How to:
Anything else on data metadata and data used. Link to data repository or explanatory article.
- Software R, minimum version 4.03
- Software R Studio, version 1.3.1073
[Describe first steps, how to use the current repository by a typical user - the digital historian with limited technical skills]
- First, work your way through LeastCostRaster_General.Rmd to follow the process step-by-step.
- Second, if you are ready to apply the batch-analysis to a region of your choice, you can run the LCPfunction.R with the calc.conductance and traveltotown functions and apply these functions to regions specified by lat/long directly. Beware that the traveltotown relies on localcity dataset to not be NULL.
- Third, go to ...
